Serial Anti–Myelin Oligodendrocyte Glycoprotein Antibody Analyses and Outcomes in Children With Demyelinating Syndromes

1Antibodies to myelin oligodendrocyte glycoprotein (MOG), detected by live cell-based assays, are found in a proportion of patients with acquired demyelinating syndromes (ADSs) of the central nervous system from early childhood through late adulthood. Since some of these MOG antibody–positive individuals will remain monophasic while others will relapse, identifying those destined for recurrent disease has important implications for both prognosis and treatment decisions. This is particularly relevant in the pediatric age group, where MOG antibodies can be found frequently (up to 40% of those with ADSs), with studies variably reporting subsequent relapsing disease in 36% to 61% of cases.,,,,,,, While it has been suggested that persistent MOG antibody seropositivity following ADS is associated with relapsing disease in both children and adults,,, the studies to date assessing serial serologic MOG antibody status may have preferentially selected patients prone to relapse, resulting in an overestimation of relapse risk. Here, we report on the long-term clinical and imaging outcomes and their association with serial serum MOG antibody measurements in a large unselected cohort of children with ADS followed prospectively from clinical onset.

2The Canadian Pediatric Demyelinating Disease Study recruited and longitudinally observed a total of 430 children presenting with a first clinical episode consistent with ADS between July 2004 and February 2017. Of these, all participants for whom a first serum sample was available within 45 days of presentation (n = 274) contributed to the current study (Figure 1A). All participants underwent comprehensive clinical assessments and were offered brain magnetic resonance imaging (MRI) scans and collection of serum samples at the time of clinical presentation; at 3, 6, and 12 months postenrollment; and annually thereafter. Clinical attacks or new MRI lesions were considered evidence of new disease activity if occurring more than 30 days from onset or 90 days in case of an acute disseminated encephalomyelitis (ADEM) presentation. Data were locked as of May 2018, and participant status was adjudicated based on information available as of their most recent examination to confer a diagnosis of either monophasic ADS (defined as the absence of new clinical attacks and of any MRI evidence of new disease activity on all serial examinations) or of multiple sclerosis (MS), neuromyelitis optica spectrum disorder, or relapsing non-MS disease, using established criteria.,, Guardians and participants provided written informed consent. Younger children provided verbal assent. The study was approved by The Hospital for Sick Children and the research ethics boards of all participating institutions.

3Clinical features at presentation were recorded, including the presence of signs and symptoms indicating optic neuritis (ON) or transverse myelitis (TM) and those meeting the criteria for ADEM. Clinical course was evaluated in terms of time from ADS onset to first relapse, annualized relapse rate over the first 4 years (computed including the presenting episode), and relapse phenotype. Disability outcome was recorded using extrapolated Expanded Disability Status Scale (EDSS) scores, as described previously. The functional system score for visual function was also evaluated. To ensure that EDSS and visual functional system scores could be compared at a consistent time from onset across all participants, we selected the scores obtained at 4 years following clinical presentation (the median postonset observation time of the cohort). Dates and durations of treatment with disease-modifying therapies were recorded.

4Myelin oligodendrocyte glycoprotein antibody status was assessed in 2022 serum samples. Of these, 5 samples obtained within 30 days of plasma exchange were excluded. All samples were tested using a live MOG IgG1-specific cell-based assay and scored semiquantitatively by a single blinded operator (Figure 1B). Any positive sample was titrated using antihuman IgG (H+L) as detecting antibody. Results of the IgG1 assay in samples obtained within 45 days from clinical onset were used to determine the serologic status at presentation. For assessment of serial samples, subsequent samples with a titer of 1:200 or greater were considered seropositive. All samples were also screened using a live cell-based aquaporin 4 (AQP4) assay as a specificity control. Where indicated, presence of cerebrospinal fluid oligoclonal bands was assessed by isoelectric focusing as part of clinical evaluation.

5Research MRI scans included axial and sagittal T2-weighted images, fluid-attenuated inversion recovery images, T1-weighted images, and T1-weighted images after administration of gadolinium. Scans acquired within 45 days of clinical presentation were analyzed by applying a standardized scoring tool modified to include additional MRI variables and blinded to MOG status and to clinical outcome.

6Serial scans were analyzed for the occurrence of new T2 lesions and lesion resolution. T2 lesions were segmented using a bayesian classifier and then manually corrected by a trained expert. T1 lesion volumes were automatically segmented as a subset of T2 lesion voxels with T1 intensity less than 87% of the normal-appearing white matter. When available, clinical scans obtained apart from the research protocol were also evaluated for the presence of new lesions.

7Figure 1A summarizes the study design. First, presenting clinical, MRI, and laboratory features were compared between anti-MOG antibody–positive and anti-MOG antibody–negative participants, as defined by the result of their baseline sample. The analysis was also repeated separately in participants older and younger than 11 years at onset and in participants with or without ADEM at presentation. We then compared the presenting clinical and MRI features of participants who were persistently positive for anti-MOG antibodies (defined as anti-MOG antibody–positive results in all serial samples up to and including at least 1 year) with those who became negative for anti-MOG antibodies. To minimize confounding of missing values and different lengths of follow-up, we conducted an additional sensitivity analysis focusing on participants for whom samples were available at all time points of baseline, 3 months, 1 year, and 2 years.

8Clinical and MRI courses were compared between participants positive and negative for anti-MOG antibodies with a minimum follow-up of 6 months (except for annualized relapse rate and EDSS, which were further restricted to participants observed for a minimum of 4 years). The analysis was repeated after excluding participants with a diagnosis of MS. The presenting features were then compared between anti-MOG antibody–positive participants with monophasic or relapsing course and used as inputs for a random forest classifier (1000 trees; Python SKLearn). A backward elimination process was applied to identify features that could work in combination to predict a relapsing outcome, with the feature with the lowest Gini impurity removed from the input at each repetition.

9Feature comparisons between groups were performed using χ2 or Fisher exact tests for categorical variables and Mann-Whitney U tests for continuous variables. The correlation between continuous variables was assessed by Spearman rank correlation. A P value less than.05 was considered statistically significant, and all P values were 2-tailed. Time to conversion to seronegative status was plotted using the Kaplan-Meier method. Analyses were performed using Python version 3.6.5 (Python Software Foundation) and R version 3.1.3 (The R Foundation).

10The presence of MOG and AQP4 antibodies was assessed in 274 children with ADS, of whom 140 (51.1%) were female, and the median (interquartile range [IQR]) age of all participants was 10.8 (6.2-13.9) years (Table 1). Myelin oligodendrocyte glycoprotein antibodies were detected in 84 participants (30.7%), while 179 (65.3%) tested negative and 11 (4.0%) had borderline results. Aquaporin 4 antibodies were detected in 1 participant, who was negative for anti-MOG antibodies. Anti-MOG antibody–positive participants were younger than anti-MOG antibody–negative participants, and 81 of 84 (96%) presented with ON, TM, ADEM, or a combination of these.

11Several clinical and MRI features differed by age at presentation. Consistent with the common presentation of ADEM, anti-MOG antibody–positive participants younger than 11 years had a high number of lesions that were often ill defined and/or distributed in a diffuse bilateral pattern, with frequent thalamic and juxtacortical involvement. Conversely, anti-MOG antibody–positive participants older than 11 years presented with fewer focal lesions, and 8 of 19 (42%) (all presenting with ON) had normal brain MRI findings at onset. When present in older anti-MOG antibody–positive participants, brain lesions more frequently showed well-defined borders. The full description of clinical, MRI, and laboratory features relative to MOG status at presentation is presented in Table 1, and a comparison between participants with and without ADEM is reported in eTable 1 in the Supplement.

12The evolution of serological status was evaluated in 216 participants with samples available for follow-up more than 1 year after baseline (median [IQR] follow-up, 4.70 [3.07-6.11] years) (Figure 2A). A total of 137 of 139 participants (98.6%) who were negative for anti-MOG antibodies at presentation remained negative in all subsequent examinations.

13Of the 67 participants who were positive for anti-MOG antibodies at onset and observed for at least 1 year from presentation, 24 (36%) remained persistently positive, 5 (7%) fluctuated between positive and negative status, and 38 (57%) became seronegative (median [IQR] time to conversion, 1.02 [0.32-1.75] years) (Figure 2B). Of 22 initially anti-MOG antibody–positive participants with ADEM with at least 1 year of follow-up, 17 (77%) became seronegative after a median (IQR) of 4.63 (3.45-12.71) months (Figure 2C). Participants who were persistently seropositive tended to be older at presentation compared with those who converted to seronegative status (median [IQR] age, 9.06 [6.60-13.36] years vs 6.95 [5.28-9.96] years; P =.03) and were more likely to have presented with monofocal ON (14 of 24 [58%] vs 9 of 38 [24%]; P =.01) (eTable 2 in the Supplement).

14Myelin oligodendrocyte glycoprotein titers showed a global decrease over follow-up that was steeper in the first year (Figure 2D). These changes in MOG antibody titers were unlikely to be influenced by treatment, as only 2 of 84 participants (2%) who were positive for anti-MOG antibodies at the time of initial presentation were ever exposed to disease-modifying therapies during their follow-up. In the sensitivity analysis of the 126 participants for whom all samples at presentation, 3 months, 1 year, and 2 years were available, the proportion of participants who converted to seronegative status was consistent with findings observed in the larger cohort (eFigure 1 in the Supplement).

15During the prospective follow-up, 66 participants met McDonald 2010 diagnostic criteria for MS, while 11 were diagnosed as having a relapsing non-MS disease, 1 as having AQP4-positive neuromyelitis optica spectrum disorder, and 196 as having a monophasic course. Among the 66 children who eventually met diagnostic criteria for MS, 11 (17%) were positive for anti-MOG antibodies at presentation. Despite meeting formal MS diagnostic criteria, all 11 of these participants had clinical and MRI features that were considered by the study clinicians to be atypical for MS (eg, clinical attacks largely restricted to ON, near-complete resolution of initial T2 lesions, no or few small lesions on follow-up scans). At the time of the most recent evaluation of these 11 patients (mean [range] follow-up from initial presentation, 8.96 [1.69-12.13] years), none of these participants were considered by their treating clinician to have a clinical disease course consistent with MS.

16Myelin oligodendrocyte glycoprotein antibodies were found in almost all participants eventually diagnosed as having relapsing non-MS disease (10 of 11 [91%]) and in 63 of 196 participants with monophasic ADS (32.1%). Of the 10 anti-MOG antibody–positive children with relapsing non-MS disease, 6 had relapsing ON, 1 had relapsing TM, and 3 had multiple neurological relapses (including either ON or TM and other neurological deficits) following initial ADEM presentation.

17Clinical relapses occurred in similar proportions of patients positive and negative for anti-MOG antibodies (53 of the latter met criteria for MS) observed for a minimum of 6 months from presentation (Table 2). When excluding participants who met diagnostic criteria for MS, the occurrence of clinical relapses was more common among anti-MOG antibody–positive participants (9 of 71 [13%] vs 2 of 121 [2%]; P =.002), ie, 10 of 11 participants with relapsing non-MS disease (91%) were positive for anti-MOG antibody at presentation. Relapses in anti-MOG antibody–positive participants were more commonly restricted to the optic nerve or spinal cord. A total of 5 of 20 anti-MOG antibody–positive participants (25%) who presented with monofocal ON and had normal findings on brain MRI had subsequent clinical relapses compared with none of the 17 anti-MOG antibody–negative children with the same presentation (P =.049). Additionally, 32 children who presented with monofocal TM had normal findings on brain MRI at onset (3 anti-MOG antibody–positive and 29 anti-MOG antibody–negative children), only 1 of whom (who was negative for anti-MOG antibodies) experienced a relapse (TM). Among participants presenting with ADEM, clinical relapses occurred in 3 of 32 participants (9%) who were positive for anti-MOG antibodies at presentation vs 1 of 30 (3%) who were seronegative. While EDSS scores were comparably low across seropositive and seronegative participants at 4 years of follow-up (Table 2), anti-MOG antibody–positive participants with a monophasic disease course had lower EDSS scores at 4 years than monophasic anti-MOG antibody–negative participants (median [IQR] score, 0 [0-1] vs 0 [0-1.5]; P =.007). This difference did not persist when the comparison was restricted to participants with the same clinical presentation of monofocal ON, monofocal TM, or ADEM.

18Serial brain MRI scans revealed that participants who were positive for anti-MOG antibodies at presentation were more likely to exhibit either complete resolution of all baseline lesions or smaller T2 and T1 total lesion volumes on their last MRI (Table 2). Among the presenting features of anti-MOG antibody–positive participants, older age was identified by the random forest analysis as the feature that was most strongly associated with a relapsing outcome, with a Gini impurity-derived importance 3-fold and 4-fold greater than the following 2 features in the ranking (T1 hypointense lesions and lesion count, respectively) (eTable 3 in the Supplement).

19Of the 16 participants who were positive for anti-MOG antibodies at presentation who experienced clinical relapses, 9 (56%) were persistently seropositive (accounting for 28% of all persistently seropositive participants), 2 (13%) had fluctuating serologic status, and 5 (31%) converted to seronegative status (Figure 3A) (eFigure 2 in the Supplement). Of the latter, 4 participants experienced at least 1 relapse after detection of the first seronegative sample without being exposed to disease-modifying therapies prior to or at the time of the relapse.

20A monophasic course was observed in most participants with an ADEM presentation, including 3 of 5 (60%) who were persistently seropositive, 27 of 28 (96%) who were seronegative, and 17 of 18 (94%) who converted to seronegative status. The 1 patient who relapsed after seroconversion had multiple relapses.

21While our protocol did not include sampling at the time of relapse, all of the samples acquired within 3 months of a clinical relapse in participants who were positive for anti-MOG antibodies at presentation were also seropositive at collection. One participant who was seropositive at presentation who then became seronegative experienced a subsequent relapse at the time of a planned study visit; findings from serum analysis at that time demonstrated re-emergence of MOG antibodies (Figure 3).

22In this prospective cohort of pediatric patients with incident ADS, we serially evaluated MOG serostatus from clinical onset, with rigorous longitudinal clinical and MRI assessments extending for up to 13 years. Several of our findings reinforce and extend prior reports, showing that MOG antibodies are common at onset in pediatric ADS,, particularly among younger children and those who present with ADEM or ON.,,, We also confirmed a bimodal distribution of clinical presentations and MRI features in anti-MOG antibody–positive children, with younger patients being more likely to present with ADEM and/or a large number of ill-defined MRI lesions and older children more often presenting with ON and no or few well-defined brain lesions.,

23A strength of our study was the procurement of serial samples at predefined intervals, enabling us to demonstrate that a negative result for MOG antibodies in proximity to the ADS presentation almost entirely excludes subsequent positive results over time. The absence of MOG antibodies at onset has key clinical implications. First, in anti-MOG antibody–negative children with isolated ON or TM (ie, with normal findings on brain MRI and no clinical features suggesting lesions in the brain) and in anti-MOG antibody–negative patients with ADEM, a monophasic outcome is highly likely. Second, anti-MOG antibody–negative patients with clinical, MRI, and cerebrospinal fluid findings meeting McDonald MS diagnostic criteria follow a typical MS disease course. The circumstance of MOG antibodies identified in children meeting MS diagnostic criteria requires specific comment. Careful review of all such patients in our cohort demonstrated a clinical and imaging disease evolution atypical of MS. As such, we agree with previous authors, advocating that the presence of MOG antibodies “plead against MS diagnosis.”

24Most patients who were seropositive at time of initial presentation subsequently became seronegative (after a median of 1 year from onset) and experienced a monophasic clinical course. Although most relapses in patients who were initially seropositive occurred while they were seropositive, nearly 40% of these patients experienced 1 or more clinical relapses after their first seronegative result. Of these, 4 of 16 participants (25%) were persistently seronegative, while 2 of 16 (13%) had anti-MOG antibodies again detected in subsequent samples. It is notable that we did not collect samples at the time of relapse and, as such, we may have missed more dynamic changes in serostatus; however, when serum was obtained within 3 months before or following a clinical attack in participants with prior seropositive status, MOG antibodies were reliably detected. Defining the optimal frequency of sampling required to more formally assess the relationship between an acute relapse and continued presence or re-emergence of MOG antibodies will require frequent regular sampling as well as sampling at time of relapse. It should also be noted that 72% of persistently seropositive patients in our cohort have not experienced a second attack despite prolonged observation. Among all presenting features of the anti-MOG antibody–positive patients evaluated in our study, an older age at presentation was most strongly associated with a relapsing outcome.

25In line with previous reports,,, most anti-MOG antibody–positive children in our cohort had a favorable outcome, with monophasic courses for more than 80% of patients, low EDSS scores at last follow-up, and either complete resolution or very low residual lesion volumes at last brain MRI scan. However, the extended long-term outcome for such patients remains to be determined.

26Based on our findings, we would advocate that MOG testing has relevance for all children with ADS. Negative testing at onset provides reassurance for families of a likely monophasic illness if their child has presented with isolated ON, TM, or ADEM. Furthermore, a positive MOG result in a pediatric patient is associated with a non-MS disease course.

27Serial serological surveillance is valuable in patients found to be seropositive at presentation, whereas it is likely to be of little clinical utility in patients identified as seronegative at presentation. Conversion to seronegative status was associated with lesser risk of subsequent relapse, although it does not entirely preclude it. In turn, persistent seropositivity is not necessarily associated with disease relapse. If serial testing following presentation is limited by resources, then a 12-month time point following initial presentation will identify more than half of the patients destined to become seronegative, including most patients with ADEM likely to convert.

28Given that many of the children found to be positive for anti-MOG antibodies at presentation will remain monophasic, we would not advocate broad institution of long-term immunomodulatory therapy following their initial presentation. In contrast, anti-MOG antibody–positive patients who do relapse are typically offered treatment with immunoglobulins, anti-CD20, or more general immunosuppressive therapies., Further research is required to confirm whether relapsing MOG demyelination is a lifelong illness and to define the morbidity of seropositive-related relapses more fully. In particular, future studies should aim to quantify the effect of seropositive-related demyelination on age-expected brain growth and white matter maturation, both of which are negatively affected in MS,,,, and on cognitive performance. All of this information is essential to better inform the risk-to-benefit ratio of long-term immunosuppression.

29This study had limitations. The study was not designed to acquire serological samples at time of relapses and might have therefore not captured transient changes in serological status associated with reoccurrence of either clinical or subclinical disease activity. Future studies, with frequent acquisition of serological samples following a first clinical episode and at the time of subsequent attacks, are needed to fully clarify the dynamic changes in serostatus in patients initially seropositive for anti-MOG antibodies. Additionally, since this study was originally designed for the study of patients with possible ADSs, patients with encephalitis-like presentations, also described in association with anti-MOG antibodies, would not have been included in our cohort.

30Anti MOG-antibodies are common in children with ADSs and are transient in approximatively half of cases. Most anti-MOG antibody–positive children experience a monophasic disease, with clinical relapses occurring more commonly, although not exclusively, in children with persistent seropositivity. Overall, most anti-MOG antibody–positive children have a favorable outcome, and the presence of anti-MOG antibodies at the time of incident demyelination should not immediately prompt the initiation of long-term immunomodulatory therapy.